Prediabetes is associated with increased risks of type 2 diabetes, cardiovascular disease, dementia, and cancer, and its prevalence is increasing worldwide. Lifestyle and pharmacological interventions in people with prediabetes can prevent the development of diabetes and possibly cardiovascular disease. However, prediabetes is a highly heterogeneous metabolic state, both with respect to its pathogenesis and prediction of disease. Improved understanding of these features and precise phenotyping of prediabetes could help to improve stratification of disease risk. In this Personal View, we focus on the extreme metabolic phenotypes of metabolically healthy obesity and metabolically unhealthy normal weight, insulin secretion failure, insulin resistance, visceral obesity, and non-alcoholic fatty liver disease. We present new analyses aimed at improving characterisation of phenotypes in lean, overweight, and obese people with prediabetes. We discuss evidence from lifestyle intervention studies to explore whether these phenotypes can also be used for individualised prediction and prevention of cardiometabolic diseases.